1. Field of the Invention
The present invention relates to a method and system for training a text-to-speech synthesis system for use in speech synthesis.
2. Introduction
Text-to-Speech (TTS) synthesis is the vehicle used to convert given text to intelligible and natural sounding speech so as to transmit information from a machine to a person. In conventional call center scenarios, TTS is only being used when and if voice talent recordings are highly impractical (e.g., address read-back) and/or too expensive (e.g., email reading in a unified messaging scenario). In all of these call-center applications, TTS voice quality (the naturalness and the intelligibility of the spoken messages) is of utmost importance. However, the resulting TTS quality for rendering conventional voice prompts and communications has been unsatisfactory.